Research and version note This version 0.1 research draft concludes the first season of Who Maintains the World? It uses work health and safety and safety reporting as examples; it is not legal advice for any Australian jurisdiction. Current local law and regulator guidance should be consulted for specific rights concerning cessation of work, representation and consultation.
Who Maintains the World? · Article 12
Quick read
Maintainers often know before executives how a system is beginning to fail. Technicians hear an abnormal sound. Care workers see needs that the formal process does not contain. Cleaning and sanitation workers encounter risks the organisation has removed from view. Frontline employees notice when a metric and reality begin to separate. Yet knowledge of failure does not by itself bring power over resources, design or purpose.
There is a legitimate reason for some separation. Maintenance knowledge is often deep and local, while a final decision may need evidence from other sites, budgets, law and public purposes. Giving any expert unlimited authority because of technical knowledge would confuse epistemic authority with political legitimacy. The opposite arrangement is also unsafe: an institution demands reliability from maintainers but withholds information, time, power to pause, and access to redesign.
Authority should be differentiated by the decision. Where an imminent and serious danger appears, people close to it need a protected route to stop or withdraw. Ordinary defects can move through reporting, investigation and temporary controls. Recurring structural faults require maintainer participation in redesign. Irreversible public decisions affecting many people need broader authorisation; a maintenance department cannot legitimately make them alone.
Australian model WHS arrangements illustrate this differentiated logic. A person conducting a business or undertaking must consult workers and health and safety representatives about hazards, risk controls and changes affecting health or safety. Properly trained HSRs have specified powers in certain circumstances, including directing cessation of unsafe work and issuing a provisional improvement notice. Rights differ across jurisdictions, but the institutional principle is clear: knowledge close to danger needs some authority capable of changing events.
Formal rights still fail when using them brings retaliation, reports disappear or contractors lack access to the same channels. Authority has to be observable in practice: concerns receive feedback, temporary controls obtain resources, and recurring faults reach redesign. Consultation held after all important options have closed is not equivalent to participation in change.
The conclusion is not that maintainers should govern the system. Authority should be proportionate to urgency, reversibility, relevance of maintenance knowledge and the breadth of impact. Maintainers must be able to see, report, receive an answer and prevent harm where necessary. Final choices about system purpose and social cost still require legitimate, accountable collective decisions.
Those who know first may be unable to act
Maintenance occurs where a system meets reality. Workers open equipment, enter service spaces, handle complaints, observe patients and reconcile records. They encounter more than quarterly averages. They see how a component wears, where a procedure fails under unusual conditions, and how much improvisation preserves the appearance of normal operation.
Management has a different view. Budgets, legal compliance, overall capacity and comparisons across sites are often concentrated there. Neither perspective is complete. Trouble begins when an organisation lets information travel upwards but keeps decisions and explanations at the top. A report disappears from the worker’s view, while the frontline repeats the same workaround.
The arrangement can make maintainers responsible for outcomes without control over conditions. A temporary repair that keeps production going is treated as evidence that the system remains viable. If failure later occurs, the worker may be asked why it was not addressed sooner. The asymmetry between responsibility and authority stays quiet until an incident exposes it.
Some workers respond by using informal power: refusing a task, warning colleagues, creating unofficial stock or bypassing a procedure. These actions may protect the service, but dependence on personal courage produces uneven safety. Formal governance should not require a person to choose between conscience and employment whenever a known weakness becomes urgent.
Maintenance knowledge does not create a general right to rule
Recognition of frontline knowledge should not become another simplistic answer. The person who best understands a machine’s condition may not know replacement priorities across an organisation. A clinician or care worker sees an individual need, while a public budget allocates among many needs. A security specialist may recommend the strongest control, but accessibility and legal rights still matter.
Expertise provides reasons about consequences, feasibility and risk. It does not alone decide which social costs should be accepted, nor does it replace the voice of people affected by the choice. Epistemic authority and legitimate decision-making intersect, but they are not identical.
Maintainers are not internally uniform either. Shifts, disciplines, employment arrangements and experience create different judgments. Long practice may hold invaluable tacit knowledge, and it may also preserve an outdated method. A system should expose experience to evidence and disagreement rather than turning proximity to the field into a new unchallengeable authority.
The question is therefore not whether one block of power should move from managers to maintainers. “Changing the system” has to be separated into different powers with different consequences.
A ladder of maintenance authority
The first power concerns information and access. Workers need risk information, maintenance history, change notices and system boundaries necessary for their task. If a supplier or management layer controls crucial evidence, the frontline carries responsibility beside a black box.
The second is a right to report, record and receive a response. Permission to submit a form is insufficient. A working route includes confidentiality or protection where appropriate, an investigation owner, a feedback period and escalation. Without a loop, the organisation collects data but does not learn.
The third is bounded operational discretion. Procedures cannot anticipate every exception. Competent workers need authority to change sequence, apply temporary controls or use modest resources within a defined range. Documentation should be proportional to risk; not every small adjustment can await distant approval.
The fourth is power to pause, withdraw or cease unsafe activity. It concerns urgent protection and is usually temporary and reviewable. A power to stop work does not allow one person to cancel a project permanently. It prevents potential harm from accumulating while the disagreement is resolved.
The fifth is participation in redesign and resource allocation. Repeated defects show that the problem is no longer a single operational deviation. Maintainers need a role in root-cause analysis, technical selection, workflow and maintainability, with visibility of why recommendations were accepted or rejected.
The sixth involves external reporting, regulatory contact or protected disclosure. When internal routes fail and serious public harm is at issue, an outside path may be the remaining safeguard. Its conditions must be grounded in applicable law; a general moral slogan cannot tell a worker what legal protection exists.
Different people and bodies can exercise these powers. An individual may withdraw immediately, an elected representative can issue a formal direction, a technical committee helps redesign, and a statutory authority decides a major change. Differentiation gives practical content to “frontline empowerment”.
Urgency and reversibility alter the proper location of power
The more urgent a decision, the greater the cost of waiting for hierarchy. When exposed electrical conductors, structural movement or a serious clinical anomaly appears, a local pause is often more reversible than continued operation. If investigation finds no danger, work can resume; harm caused by continuation may not be recoverable.
Closing an entire public service, replacing a technology permanently or reallocating a large budget affects more people. It requires broader evidence and authority. Field knowledge remains necessary, but does not complete the final decision.
A useful principle follows: as risk becomes imminent, temporary protective authority should move closer to the field; as action becomes less reversible and its effects spread, decision-making should combine more forms of knowledge and formal legitimacy. This is not a numerical formula. An institution has to specify authority before a crisis rather than asking workers to improvise it during one.
Reversibility must also be judged from the position of affected people. Pausing a machine imposes cost but usually permits restoration. Continuing a faulty eligibility system for another day may produce losses that are difficult for many individuals to remedy. Management should not classify only organisational expenditure as irreversible while assuming personal rights can be repaired later.
Uncertainty does not always point towards continuation. In a high-consequence setting, a short pause may preserve options while evidence is gathered. In other contexts, interruption itself can be dangerous. The allocation must be designed around the particular service and failure modes.
Work health and safety provides an institutional example
Safe Work Australia’s material on consultation states that PCBUs must consult workers and their HSRs when identifying hazards, assessing risks, deciding controls and proposing changes that may affect health or safety. Consultation is not merely courteous. Workers have direct knowledge of work and its risks.
Under the model WHS arrangements, an HSR represents a work group, participates in health and safety matters and can access certain information about hazards. Safe Work Australia’s HSR guidance explains that a trained HSR can, in specified circumstances, direct workers to cease unsafe work or issue a provisional improvement notice. Adoption and detailed procedure differ among states and territories, so model-law information is not uniform local legal advice.
The International Labour Organization’s Convention No. 155 contains a related principle: a worker who removes themself from a work situation they reasonably believe presents imminent and serious danger to life or health should be protected from undue consequences according to national conditions and practice.
These arrangements do not transfer all enterprise decisions to maintenance workers. They recognise that some knowledge will arrive too late to influence action if it carries only a right of expression and no protection or temporary restraining effect.
The details matter. Training, representation, issue-resolution procedures and regulator access turn an abstract right into something a worker can use. A rule without awareness or support may exist legally but remain absent from daily operations.
A reporting system can become an information cemetery
An organisation may own an advanced work-order or incident platform without maintaining feedback. A worker enters an anomaly, the software generates a number, and no investigation conclusion, interim control or timetable returns. Eventually, low reporting is interpreted as improvement when it may mean people no longer expect reporting to help.
Aviation safety culture offers principles worth examining, though not mechanically transplanting. ICAO’s Safety Culture material explains that willingness to report experience and error depends on perceived benefits and disadvantages, and that trust, support and fair treatment shape reporting culture. ICAO material on frontline reporting and feedback adds that people are more likely to report when they receive meaningful feedback and observe action; frontline personnel should not be reduced to data providers.
A just culture is not a no-blame culture. Inadvertent error, system conditions, gross negligence and wilful violation require distinctions. Punishing every mistake destroys weak signals. Exempting every conduct from accountability destroys the boundary of safe practice. Fair judgment needs known standards, independence and review rather than retrospective managerial convenience.
Feedback has epistemic value as well as motivational value. A maintainer can identify when investigators misunderstood the field or proposed a control that cannot be performed. Analysis that remains permanently at headquarters refuses an opportunity to correct the institution’s model of its own work.
Report counts should therefore be treated cautiously. An increase may reflect deteriorating conditions or growing trust. A decrease may reflect safety or silence. Evaluation needs evidence about investigation quality, action, time, recurrence and reporter experience.
Outsourcing can sever knowledge from authority
Contract workers may be closest to equipment, sanitation, care, software or facilities while remaining outside the client organisation’s routine governance. A contract defines them as service providers, and a field problem travels through their employer, a principal contractor and the client.
Each layer can compress the signal. Reporting a defect may harm a service metric. A supplier fearing contract renewal may delay escalation, while the client assumes the contract has transferred the risk. Formal allocation of liability has not created practical authority at the site.
A more reliable arrangement includes contract maintainers in relevant information, change consultation, incident channels and operational meetings. It states who can pause, who funds temporary measures and how a reporter is protected. Shared duties across organisations need rehearsed paths, not only clauses in an annex.
Procurement should evaluate maintainability as well as acquisition price. A system accessible only to the original vendor, without sufficient documentation or with long spare-part lead times, creates future dependence. Maintainer participation can expose these costs before design and contract decisions lock them in.
The issue is not solved by insourcing everything. Specialist suppliers can hold valuable knowledge and capacity. The governance task is to prevent organisational boundaries from becoming places where signals, records and responsibilities disappear.
Design participation is not a veto against innovation
Designers can treat maintenance as an after-launch concern. Yet equipment access, software logging, consumable standards, data export and manual takeover are determined during design. Later maintenance difficulty is often the delayed consequence of an early choice.
Maintainers can test assumptions against work. Can a component be replaced safely? Can the night shift obtain support? Do alarms distinguish urgency? Will records remain readable after supplier exit? These are life-cycle questions, not simply conservative resistance.
Maintenance advice still needs coordination with other purposes. The easiest system to maintain may not be the most energy-efficient, accessible or useful. Participation makes future work and risk present in today’s choice; it does not give one department a veto over every change.
Where a proposal is rejected, decision-makers should record why and what risk they accept. Transparent disagreement respects knowledge more than an invitation followed by silence. It also provides evidence for later evaluation.
Participation should begin early enough to alter options. Asking maintainers to comment after procurement has selected a product converts involvement into an implementation briefing. Authority is partly temporal: a voice that arrives only after irreversibility is not the same voice.
Users and the public hold knowledge that maintainers may not
Maintainers may identify technical failure without seeing every affected person. An efficient service can exclude a disabled user. A data team can identify performance degradation but miss the procedural burden experienced by an appellant. Maintenance governance needs user experience and public value as well as operational knowledge.
Major change usually crosses several layers. Experts describe feasibility, maintainers explain conditions of continuation, users show lived consequences, executives allocate resources, and authorised public institutions settle public purposes. None possesses the whole.
This limits the appeal of “let experts decide”. Experts deserve authority on relevant factual questions, while conflicts of value and rights need broader legitimacy. Democratic authority, conversely, does not make technical facts optional. A good institution lets these forms of authority constrain one another without allowing either to absorb the other.
Public transparency matters when the maintained system allocates rights or common resources. Confidential internal reporting can protect workers and safety information, but the institution still owes an aggregate account of known risks, responses and unresolved uncertainty. Protection and accountability must be designed together.
Maintenance labour must itself be maintainable
Responsibilities without time and staffing turn participation into extra work. HSRs, quality workers, open-source maintainers, carers and field technicians may be expected to manage systemic risk outside normal duties. The institution then depends on personal sacrifice.
Maintainers require training, relief, physical and psychological safety, stable scheduling, and protection from disadvantage for reporting. Continuous on-call work and repeated emergency response degrade judgment. A system that remains reliable by exhausting maintainers has merely relocated its deterioration into human bodies.
Authority needs support as well. A stop-work decision may attract intense pressure, protected disclosure carries career risk, and design participation requires access to evidence. Formal text is difficult for an individual to use without representatives, unions, independent regulators, legal paths or management commitment.
The maintainer group also needs renewal. Apprenticeship, handover and cross-role learning prevent power and knowledge from accumulating in a few irreplaceable people. Sharing knowledge does not devalue expertise. It allows the system to continue without consuming an individual.
This is an important limit on heroic narratives. Praising workers for keeping the world going can become a way of normalising conditions in which they are never allowed to stop. Recognition without resources or authority is another form of invisibility.
A Sustenesis reading of maintenance and change
In this series’ applied framework, maintenance does not hold a structure forever in its current form. A system sustains relative continuity through repeated correction, replacement, interpretation and coordination. Maintainers work at the boundary between continuity and difference. They learn which changes can be absorbed and which are destroying the conditions of continued operation.
That knowledge makes maintainers part of institutional knowing, but not its single subject. A Sustenesis perspective emphasises the distributed structure: management, professional experts, field workers and affected people each encounter different constraints. No position has the whole.
Authority design should allow those partial forms of knowledge to alter action at the appropriate time. If only the centre may change anything, the structure loses sensitivity to local difference. If every local part can rewrite the whole, common commitment becomes impossible. Stability in Sustenesis is not the suppression of change; it depends on relations able to absorb, judge and take responsibility for change.
Maintenance authority is therefore neither a reward for effort nor a concession from management. It is one of the structural means by which a system remains capable of knowing its own condition. Its limits are equally structural because local knowledge alone cannot establish every purpose.
Current judgment: maintenance knowledge needs proportionate power
Maintainers should neither rule by virtue of expertise nor be reduced to executing predetermined instructions. Appropriate authority depends on four considerations: urgency of risk, reversibility of action, relevance of maintenance knowledge, and breadth of the decision’s effects.
At minimum, institutions should guarantee necessary information, protected reporting, defined feedback and escalation. Imminent serious danger requires a practical route to pause, withdraw or cease. Recurring problems must enter root-cause analysis, with maintainers and affected users involved in design and resource discussion. Broad and irreversible public choices still require lawful authority and public reasons.
These arrangements need audit. An organisation should examine not only report volume but whether reports were investigated, answered and acted upon, and whether contractors and precarious workers can use the route on equal terms. A power written in policy but punished in use does not operate as power.
The version 0.1 conclusion is that the world depends on maintainers, but it cannot depend only on their goodwill, endurance and improvisation. A sustainable institution allows the people who first see a crack to prevent its expansion without placing the direction of the shared world on them alone. Maintenance becomes a public capacity where knowledge, authority and responsibility are reconnected.
Primary sources and further reading
- Safe Work Australia, Consultation.
- Safe Work Australia, Health and safety representatives and work groups.
- International Labour Organization, Occupational Safety and Health Convention, 1981 (No. 155).
- International Civil Aviation Organization, Safety Management Manual: Safety Culture.
- ICAO, Frontline reporting and the role of feedback.
Series navigation: Who Maintains the World? series overview
Discover more from Geoffrey Chen
Subscribe to get the latest posts sent to your email.